Catching the mean chocolate thief with Raspberry Pi, Bluemix, and Cloudant

I always try to have some chocolate in my office, kind of as mood enhancer. But how to be sure that nobody else is going to plunder and pilfer my hidden treasures? So it was great that last week at the Developer Week conference in Nuremberg I got my hands on a Raspberry Pi (thank you, Franzis Verlag and Christian Immler) and that I know a little about IBM Bluemix. And here is the plan: Hook up my IBM-sponsored webcam to the RPi and then take, activated by a motion-sensor, a snapshot and upload the picture and metadata to a Cloudant NoSQL database. With a Bluemix-based application I could then have worldwide access to the "incident data" and catch the mean chocolate thief...

Raspberry Pi, motion sensor, and webcam

The first step is the hardware setup. Connecting the pins of infrared motion sensor to 5V, ground, and an IO port on the Rasperry, and then the webcam to the USB port is simple. The mini-computer already has LAN access which is important to access the Cloud services.

Next I logged into IBM Bluemix, the platform-as-a-service (PaaS) offering for developers and created a Cloudant data store. This is done similar to how I described it in my previous article on using Cloudant for some statistics for a weather webpage. The account data for the Cloudant database can be obtained in JSON format. I copied that information into a file "cloudant.json" and placed it into my project directory on the Raspberry Pi. With that, we are already at the software part of this project.

In the following, you see the Python script I used for the prototyping. It is performing some setup work which includes reading in the access information for the Cloudant account. The main part is a simple loop waiting for the thief to appear, i.e., the motion sensor to be actived:

 import datetime  
 import time  
 import subprocess  
 import RPi.GPIO as io  
 import json  
 import couchdb  
 io.setmode(io.BCM)  
   
 pir_pin = 18  
 scriptPath='/home/pi/projects/officeCam/takeSnap.sh'  
 imgFile='/home/pi/projects/officeCam/office.jpg'  
   
 # couchDB/Cloudant-related global variables  
 couchInfo=''  
 couchServer=''  
 couch=''  
   
 with open("cloudant.json") as confFile:  
   couchInfo=json.load(confFile)['cloudantNoSQLDB'][0]  
   couchServer=couchInfo["credentials"]["url"]  
   couch = couchdb.Server(couchServer)  
   
 # access the database which was created separately  
 db = couch['officecam']  
   
 io.setup(pir_pin, io.IN) # activate input  
   
 while True:  
   if io.input(pir_pin):  
     subprocess.call([scriptPath])  
     f=open(imgFile,'r')  
     # basic doc structure  
     doc= { "type" : "oc",  
        "creater" : "RPi",  
        "location" : "office",  
        "city" : "Friedrichshafen"  
       }  
     doc["timestamp"]=str(datetime.datetime.utcnow())  
     # and store the document  
     db.save (doc)  
     db.put_attachment(doc,f,filename='cam.jpg')  
     f.close()  
   
     print("Alarm processed")  
   time.sleep(1)  
   

Once some motion has been  detected, the Python script invokes a shell script. It is printed below. The only action is to execute the fswebcam program which takes a snapshot with the webcam. Thereafter, back in Python, I create a JSON document, stuff the current timestamp and some other information into it and store it to the Cloud-based NoSQL database. As last step I attach the picture to that document, so that even if the mean chocolate thief notices the trap, the image is secured in the cloud.

   
 #!/bin/sh  
 fswebcam -q -c /home/pi/projects/officeCam/fswebcam.conf  

With that I am done with the Raspberry Pi. What is left is to work on the reporting. See how it is done in Python on Bluemix and Cloudant.